Accuracy of an artificial intelligence-enabled diagnostic assistance device in recognising normal chest radiographs: A service evaluation

AI demonstrated high reliability in identifying normal CXRs, with a negative predictive value of 99.43% and an AUROC of 91.92%. Sensitivity and specificity were 96.88% and 75.55%, respectively. Sensitivity analysis showed NPV remained high (88.96%–99.54%) across varying prevalence levels. The AI device recognized normal CXRs with high NPV and has the potential to increase radiologists’ productivity.